Generational Dynamics
Fourth Turning Forum Archive


Popular links:
Generational Dynamics Web Site
Generational Dynamics Forum
Fourth Turning Archive home page
New Fourth Turning Forum

Thread: Global Warming - Page 163







Post#4051 at 12-23-2013 09:46 PM by The Grey Badger [at Albuquerque, NM joined Sep 2001 #posts 8,876]
---
12-23-2013, 09:46 PM #4051
Join Date
Sep 2001
Location
Albuquerque, NM
Posts
8,876

Quote Originally Posted by Deb C View Post
Just curious as to why you think she is unlikely to be quotable. Is it because she , as you indicate, will be exiled from her position like Churchill?
My guess is, because her style is wonkish and matter-of-fact, unlike the fire-eating Churchill.
How to spot a shill, by John Michael Greer: "What you watch for is (a) a brand new commenter who (b) has nothing to say about the topic under discussion but (c) trots out a smoothly written opinion piece that (d) hits all the standard talking points currently being used by a specific political or corporate interest, while (e) avoiding any other points anyone else has made on that subject."

"If the shoe fits..." The Grey Badger.







Post#4052 at 12-23-2013 09:56 PM by Deb C [at joined Aug 2004 #posts 6,099]
---
12-23-2013, 09:56 PM #4052
Join Date
Aug 2004
Posts
6,099

Unlike so many of us older folks, the youth have a can do attitude. May their courage be contagious.

"When student leaders held the first Divestment Convergence last year on Swarthmore's campus – I remember seeing this sign and it sticking with me: “Divestment is just a tactic – Climate Justice is the goal.”"


All I Want for Christmas: More Youth-Led Climate Justice

A holiday vision for global activism in 2014


Young college students today have far more power than you realize. You are the first generation that isn't just fighting for your grandkids, or for your children per say – but for your own future. It's your future on the line. Your very tomorrow is being fracked. And that gives you tremendous power as organizers—to use your voice of moral urgency.

And we need that power, and your voices. We need more youth-led tactics, strategies, and visions. We need you to rise far beyond the scope of divestment activists in 2014, and as full leaders in this work.

To be clear, we are in real trouble. We are in borrowed time. The people of New York's Rockaways and the Philippines can tell you this. Super storms like Sandy and Haiyan are here. The threat is here. Our world is living in fear—and a state of unraveling.
So, let me tell you what else I see when I shut my eyes.

When I shut my eyes for the new year:

In 2014, I imagine students rising up like never before to challenge fracking around this country. I imagine days of action in 2014 when divestment activists take treks off their campus to local fracking wells and gas plants, and use their bodies to shut down 100 wells and plants in one day. And then start to hold them. The only threat as everywhere as the threat to clean endowments is the threat to clean water—and chances are that threat is alive and well not far from your campus walls. You can learn more about it this March at the first Shale Field Justice Spring Break.


Would we be willing to follow their lead?

In 2014, I imagine students moving not just money—but the hearts and minds of America. It is the year that this country's people realize that it's their youth who are being truly truly fracked, and their youth won't stand it anymore. You have lit this country on a different kind of fire. And the people themselves start to join this early rebellion.

To be clear, you have the power. You and frontline activists, and Indigenous leaders. And for one, I am willing to follow your lead in 2014.


If you are a weathered divestment activist—if you've been involved for a couple semesters and been a part of the great anti-oppression and solidarity work that has bled into so many campuses, then you are called to be more than a divestment activist in 2014. You are called to be a guardian of the youth climate justice movement – you are called to be a leader.


What is your vision for 2014?


http://www.commondreams.org/view/2013/12/23-6


"We need youth power far more than ever before."
Last edited by Deb C; 12-23-2013 at 10:02 PM.
"The only Good America is a Just America." .... pbrower2a







Post#4053 at 12-24-2013 03:30 AM by Vandal-72 [at Idaho joined Jul 2012 #posts 1,101]
---
12-24-2013, 03:30 AM #4053
Join Date
Jul 2012
Location
Idaho
Posts
1,101

Quote Originally Posted by Mikebert View Post
Support you did the group analysis you prefer and got a significant result. Then the next six groups show no significant result. This can happen too. This is why a p value is assigned to it.
The confidence in detecting an effect from a proper study of storm distributions will always be higher than a conclusion drawn on a single event. The single event tells you nothing about rejecting the null hypothesis.

No. But what if the recipient of the hand had announced immediately before that he was going to get a royal flush?
What does that have to do with our discussion?

You wrote “A 20% bias (heads = 60%) will not require that many but it will require more than 100 flips. It's actually 400 flips for a z-score of four sigma confidence.”

What did you use for degrees of freedom in this calculation? I suspect none. From Wiki:
From the appropriate Wiki article: n = Z^2 / 4 E^2

In the example above you are doing probability analysis, the first bolded analysis. Degrees of freedom do not apply because you know the parameters of the coin flip population ahead of time. Degrees of freedom arise when you are inferring population parameters from a sample.
Which you can not possibly know for all possible storm wind speeds!

The 4 sigma analysis is also a probability analysis. The null hypothesis (the one being tested) holds that the perturbation has no impact. Therefore storms afterward fall into the same population as those before. The "before" population can be analyzed with N-1 degrees of freedom (N = # historical storms) to estimate its parameters. Given the parameters, the probability of the super storm occurring can then be estimated. If the probability is low then the null hypothesis is not supported. That is, it is unlikely that the “before” distribution is valid for the “after” situation.
If a single superstorm occurs, you can only conclude that something that was unlikely but possible, did in fact happen. You can not link that rare event to warming. If you get several superstorms, meaning the storm distribution is significantly different, you can statistically conclude that there is something different now.

Now in order to do this you do need a large sample of “before” data in order to have a large degree of freedom) so that the sample parameters closely approximate the population parameters.
Fine, but a large enough sample will include several four sigma storms, pre-warming, showing you that warming is not necessary for them to occur! So if you observe a single four sigma later you can not attribute its existence to warming.

In many cases you do not have a pre-existing reference population and so have to estimate the parameters for the reference population from a small sample of control experiments. But in this case, as in quality control, you have a good idea of reference population parameters. If you don’t, then you cannot define a four sigma event (remember sigma the population standard deviation, not s the sample standard deviation). The assumption made by the term “four sigma storm” assumes that one can do this. In reality it might not be the case. But you made an assertion that for all reference populations (even ones which are exactly known like coins or cards) one can never infer valid probabilities for single events occurring over a pre-specified period of time.
Not quite what I am arguing. I am pointing out that relying on a single event does not allow you to posit "warming effect" as your alternative hypothesis. A single event may be very improbable but if it occurs all you can say is that something rare happened, you can not attribute cause to the event. If several very improbable events occur in a short time period you can conclude that the probabilities have been altered. The changed probabilities require an explanation. A single event can not show you that the probabilities are altered!

I am saying that is not true.
So. I am trying to show you that calculating how unlikely an observed event is can not tell you that anything is actually amiss. Rare events do happen!

Statistics is the mathematical body of science that pertains to the collection, analysis, interpretation or explanation, and presentation of data. Significance testing is one aspect of statistical analysis that estimates the probability of outcomes. It is most closely akin to probability analysis.
You can't do probability analysis of storms. They are not discrete phenomena.







Post#4054 at 12-24-2013 09:29 AM by Mikebert [at Kalamazoo MI joined Jul 2001 #posts 4,502]
---
12-24-2013, 09:29 AM #4054
Join Date
Jul 2001
Location
Kalamazoo MI
Posts
4,502

Quote Originally Posted by Vandal-72 View Post
The confidence in detecting an effect from a proper study of storm distributions will always be higher than a conclusion drawn on a single event.
Why?

What does that have to do with our discussion?
Are you saying there is no difference between someone being dealt a royal flush out of the blue and one getting dealt one immediately after expressing the intent to get one?

From the appropriate Wiki article: n = Z^2 / 4 E^2
And did you use degress of freedom to come up with that p value? Why not?

Which you can not possibly know for all possible storm wind speeds!
Which you do not know in general. If you run an experiment and perform an assay to give you a result, the number of possible outcomes will depend, in part, on the precision of your measurement. As precision rises the number rises with it. In general you do not even know the number of possible outcomes, much less what they are.

If a single superstorm occurs, you can only conclude that something that was unlikely but possible, did in fact happen. You can not link that rare event to warming.
You can never prove an hypothesis, only reject them. One can give supporting evidence for an hypothesis by showing the null case is rejected. And this can be done with any number of experimental datapoints, including a single one--provided you do it right.

If you get several superstorms, meaning the storm distribution is significantly different, you can statistically conclude that there is something different now.
If a single event with probability p is meaningless why are multiple events with a joint probability of p meaningful? And don't just assert that the larger number confers some sort of mystical meaning. Explain why you believe this.

Fine, but a large enough sample will include several four sigma storms, pre-warming, showing you that warming is not necessary for them to occur!
Who said warming was necessary? There is a world of difference between "warming makes superstorms more likley" and "warming makes superstorms possible".

So if you observe a single four sigma later you can not attribute its existence to warming.
You keep saying this, but you give no reason why you believe this. At one point you tried to relate it to degress of freedom.

Now you are using this notion:

You can't do probability analysis of storms. They are not discrete phenomena.
Well the strong storms get names, are assigned begining and endings. They seem to me like other historical events/periods such as "War of 1812", "Election of 2000", Summer of 1967 or Hurricaine Camille. They can be counted. In what relevant way are they nondiscrete?
Last edited by Mikebert; 12-24-2013 at 10:06 AM.







Post#4055 at 12-24-2013 09:37 AM by B Butler [at joined Nov 2011 #posts 2,329]
---
12-24-2013, 09:37 AM #4055
Join Date
Nov 2011
Posts
2,329

Left Arrow He just sets a high bar...

Quote Originally Posted by Deb C View Post
Just curious as to why you think she is unlikely to be quotable. Is it because she , as you indicate, will be exiled from her position like Churchill?
I don't think she'll be unquotable. I just don't think she's apt to be as quotable as Churchill. In my book, the only person as quotable as Churchill is Shakespeare. Churchill would work at it. To hear some tell it, he'd wake up in the morning and spend some time coming up with an immortal famous quotation, then spend the day trying to find or create circumstances appropriate to use it.

Find a reference book on famous English language quotes and you'll see what I mean. Most everyone ought to know of the several famous speeches he made during Britain's Finest Hour. His mastery of the English language isn't limited to those speeches.

I may be prejudiced. In an on line role playing game, I once played a character who was a big fan of Churchill. I kept my book of Churchill quotes by my computer at all times. A necessary tool. Most weeks, I would preface a short story fragment intended to advance the game's plot with a Churchill quote. I seldom had difficulty finding an appropriate quote.

However positive one might feel towards Elizabeth Warren, a book on her wit and wisdom has yet to be published.

Will she be exiled from her position? Can't answer that one for sure. Senators get in in six year chunks. She is running out of Massachusetts. How aggressively will she stick to her guns? Is she deliberately setting herself up for a Grey Champion like role, accepting short term hits to stake a long term position, or is she fighting for principle on principle?

I can't see clearly that far downstream, but I'll give her a hearty "You go girl!"
Last edited by B Butler; 12-24-2013 at 10:04 AM.







Post#4056 at 12-24-2013 12:19 PM by Deb C [at joined Aug 2004 #posts 6,099]
---
12-24-2013, 12:19 PM #4056
Join Date
Aug 2004
Posts
6,099

Quote Originally Posted by B Butler View Post
I don't think she'll be unquotable. I just don't think she's apt to be as quotable as Churchill. In my book, the only person as quotable as Churchill is Shakespeare. Churchill would work at it. To hear some tell it, he'd wake up in the morning and spend some time coming up with an immortal famous quotation, then spend the day trying to find or create circumstances appropriate to use it.

Find a reference book on famous English language quotes and you'll see what I mean. Most everyone ought to know of the several famous speeches he made during Britain's Finest Hour. His mastery of the English language isn't limited to those speeches.

I may be prejudiced. In an on line role playing game, I once played a character who was a big fan of Churchill. I kept my book of Churchill quotes by my computer at all times. A necessary tool. Most weeks, I would preface a short story fragment intended to advance the game's plot with a Churchill quote. I seldom had difficulty finding an appropriate quote.

However positive one might feel towards Elizabeth Warren, a book on her wit and wisdom has yet to be published.

Will she be exiled from her position? Can't answer that one for sure. Senators get in in six year chunks. She is running out of Massachusetts. How aggressively will she stick to her guns? Is she deliberately setting herself up for a Grey Champion like role, accepting short term hits to stake a long term position, or is she fighting for principle on principle?

I can't see clearly that far downstream, but I'll give her a hearty "You go girl!"
Thanks for the education in regards to Churchill. I wasn't aware of his mission to be quotable or the energy that he put into being so.
"The only Good America is a Just America." .... pbrower2a







Post#4057 at 12-24-2013 06:37 PM by nakile [at joined Jun 2013 #posts 48]
---
12-24-2013, 06:37 PM #4057
Join Date
Jun 2013
Posts
48

To get this topic back on track, tell me I'm not the only one who hasn't noticed nuclear fusion has been quickly sneaking up toward viability? Vices Motherboard wrote a good summary.

General Fusion is working on a 100 MWe deuterium-tritium reactor. Their method uses mechanical shockwaves shot into a vortex inside a core of spinning molten lead to provide the compression needed to heat the fuel to break-even temperatures. That very same spinning core of lead blocks the radiation, breeds the tritium needed, and transfers the heat to a steam loop. They think that in terms of engineering they have the simplest approach and hope to have a have a proof of concept in 2014. They're funded by private investors and the Canadian government.

Lockheed-Martins Skunkworks is working on a 100 MWe high beta fusion reactor also using deuterium-tritium. Their goal is a reactor that will fit on the back of semi-truck for transport and hope for a commercial prototype by 2017. Not much else has been said, but there's speculation this an evolution of Robert Bussards polywell design. Here's a video presentation by the project leader.

LPP is working on proton-boron-11 fusion. This a big deal if they can get it to work: when you fuse boron-11 with a proton you get an aneutronic reaction, that means you get a clean nuclear split with no radiation and in the case of pB11, the waste product is helium. There are a few radioactive side reactions, but they're easily managed. The helium atoms are also electrically charge, which is then captured and used directly without the need for a steam turbine (you would probably need an inverter, though). They're aiming for a 5 MWe reactor the size of a garage.

There's a few others too, but I have a post limit right now so I don't think I could fit them.

All of these projects are promising proof of concept within 2-5 years, which is much better than 30. I think it's time that we begin to at least entertain the idea of nuclear fusion being a solution to global warming. Build these next to–or in–cities and not only use the electricity, but pipe the waste heat in and provide a district heating service, maybe even use absorption chillers for a cooling service. Then mass produce the entire system and pay for it using long term no interest government loans and create municipal level energy providers. Then sell it overseas and the global warming crisis is basically over.







Post#4058 at 12-25-2013 02:09 AM by Vandal-72 [at Idaho joined Jul 2012 #posts 1,101]
---
12-25-2013, 02:09 AM #4058
Join Date
Jul 2012
Location
Idaho
Posts
1,101

Quote Originally Posted by Mikebert View Post
Why?
Because a single event, no matter how unlikely, can absolutely be due to random chance. A shift in the distribution pattern for all storms can not be due to random chance.

Are you saying there is no difference between someone being dealt a royal flush out of the blue and one getting dealt one immediately after expressing the intent to get one?
Statistically? Yeah, there is no difference. Your search for a four sigma storm is not the same as a search for a crooked dealer. A single dealt hand is not evidence of anything.

And did you use degress of freedom to come up with that p value? Why not?
Did you even read the article? That wasn't a p-value calculation. It was a tool for determining the number of flips required to statistically conclude that a coin is 60/40 rather than 50/50 at a z number that YOU CHOSE.

Which you do not know in general. If you run an experiment and perform an assay to give you a result, the number of possible outcomes will depend, in part, on the precision of your measurement. As precision rises the number rises with it. In general you do not even know the number of possible outcomes, much less what they are.
Exactly. That's why your insistence on "probability analysis" for a single storm is ridiculous.

You can never prove an hypothesis, only reject them.
Speaking statistically, you reject the null hypothesis and accept, not prove, the alternative.

One can give supporting evidence for an hypothesis by showing the null case is rejected. And this can be done with any number of experimental datapoints, including a single one--provided you do it right.
No you can't. A single point is not a dataset.

If a single event with probability p is meaningless why are multiple events with a joint probability of p meaningful?
Because they indicate that your initial probabilities are likely not valid anymore. Something has changed! That's the whole point. If you get enough data points showing that storm strength distribution is different from what it used to be, you can begin to accept the alternative hypothesis that warming has changed storm patterns.

Two big storms by themselves is very unlikely to be enough data to reject the null but two big storms within a shifted set of more typical storms might be. A single storm stripped of the context of the rest of the storms can never be a sign of statistical significance.

And don't just assert that the larger number confers some sort of mystical meaning. Explain why you believe this.
It's not my belief. It's how statistical analysis is actually done.

Who said warming was necessary? There is a world of difference between "warming makes superstorms more likley" and "warming makes superstorms possible".
But your four sigma storm test can not possibly support the second statement. You assigned a storm's sigma value based upon the pre-warming dataset, which directly means that non-warming makes such storms possible!

You keep saying this, but you give no reason why you believe this. At one point you tried to relate it to degress of freedom.
It's not a belief. It's how statistics is actually done. You tried pretending that a certain statistical test of a single data point was valid (Grubb's test). Care to explain why you dropped that claim? Care to tell us what statistical test for significance you are relying on now? I've gone through the list of typical statistical tests for significance and none of them allow data sets of a single point because the degrees of freedom typically take the sample size minus one or two. That makes the statistical calculation rather difficult because it will require you to divide by zero. But, if you are claiming to have figured out how to do that, then by all means share!

Now you are using this notion:

Well the strong storms get names, are assigned begining and endings. They seem to me like other historical events/periods such as "War of 1812", "Election of 2000", Summer of 1967 or Hurricaine Camille. They can be counted. In what relevant way are they nondiscrete?
Because you said wind speed! Your four sigma value is a measure of wind speed! Do you really not grasp the difference between continuous and categorical?

At first, I thought maybe you were describing some statistical technique that I was unfamiliar with. But, the more responses you write the more I'm convinced that you really don't get how basic statistical principles work.







Post#4059 at 12-25-2013 10:43 AM by B Butler [at joined Nov 2011 #posts 2,329]
---
12-25-2013, 10:43 AM #4059
Join Date
Nov 2011
Posts
2,329

Left Arrow Ivory Towers vs Factories?

Mike

I have learned that when you join a technical or scientific discussion, it is generally prudent for me to get out of your way. You generally are more persistent and have a broader experience than I. None the less, I’ll make a few comments.

In a religious discussion centered on how one best does God’s will, one might see various perspectives advocated: faith, good works, following revealed texts, heeding moral rules and simply loving one’s neighbor might stand as examples. Different traditions might emphasize some over others. I would expect that if one examined the origins and major crises associated with each tradition, one might often find good justifications for the differences in doctrine and emphasis.

Understanding the origins of the differences, however, wouldn’t make me expect it to become easier to achieve harmonious agreement.

I’m trying to figure out a dividing line regarding the current statistical ‘discussion.’ There ought to be more happening than one of you being ignorant or stupid while the other is a shining example properly placed on a hill. The closest I can come up with is that one of you is a ivory tower academic preserving the purity of his methods while the other lives in the real world and has to use statistical methods to solve real and practical problems.

In one of your examples, the long string of numbers was associated with a factory process. One might assume that if the number associated with a given batch is within a certain distribution, that batch is ‘sellable,’ a rather important thing to know in a real world environment. If the number goes outside that boundary, does one want to analyze that batch run to find out why? Does one want to improve the production process to reduce the chance of a bad run? If economics is part of the picture, do other things become important? What is the cost per batch? How often does one get a ‘bad’ batch? How much might it cost to determine why a batch went bad, or to improve the production process?

And what is the culture of the company? I would not think every company would answer such questions in the same way.

I can see that those who live in academic ivory towers — do high schools have ivory towers? — might be more concerned with the purity of definitions and nuances of language, more so than a vice president in charge of quality control. The VP is just looking for good advice on how to improve his product or improve his bottom line.

Do you have other insight on the core perspective differences beneath the discussion?

Bob







Post#4060 at 12-26-2013 01:56 AM by Bad Dog [at joined Dec 2012 #posts 2,156]
---
12-26-2013, 01:56 AM #4060
Join Date
Dec 2012
Posts
2,156

Do I need to direct y'all to the cash vault, and buy as many quarters as possible?







Post#4061 at 12-26-2013 09:22 AM by Mikebert [at Kalamazoo MI joined Jul 2001 #posts 4,502]
---
12-26-2013, 09:22 AM #4061
Join Date
Jul 2001
Location
Kalamazoo MI
Posts
4,502

Quote Originally Posted by Vandal-72 View Post
Because a single event, no matter how unlikely, can absolutely be due to random chance.
And so can multiple events.

Statistically? Yeah, there is no difference.
What? Of course there is a difference. The probability of a being dealt a royal flush in N hands is given by

probability = 1 - (1-p)^N

Where p is 1/649740 and N is the number of hands being considered. When a person announces he is going to get a royal flush on the next deal he is saying it will occur in that hand and that hand only. N = 1 and the probability is p. Since p is very small it is very likley that a special cause is involved and I would strongly suspect the guy is cheating. I cannot believe you would hold no suspicions.

Now if someone gets dealt a royal flush out of the blue the exact same formula applies. But what is N? In this case no constraints that put an upper limit on N have been supplied (no predictions were made). So N is going to be the number of all poker hands that have been or will be dealt, because the set of hands under consideration was not specified. That is, N is going to be a really big number. For large N the expression above evaluates to 1. So the the probability is essentially 100%. In the first case, the null hypothesis is rejected, in the second it is not.

Did you even read the article? That wasn't a p-value calculation. It was a tool for determining the number of flips required to statistically conclude that a coin is 60/40 rather than 50/50 at a z number that YOU CHOSE.
I read the article and I saw no mention of degress of freedom (DoF). Did you?

I've gone through the list of typical statistical tests for significance and none of them allow data sets of a single point because the degrees of freedom typically take the sample size minus one or two.
And this is the issue. You apply formulas. You do not understand what you are doing when you apply these formulas. So you are making an argument from authority.

Because you said wind speed! Your four sigma value is a measure of wind speed! Do you really not grasp the difference between continuous and categorical?
Maximum sustained wind speed is not a continuous variable: there is one value per storm. It is explicitly used to categorize storms.

At first, I thought maybe you were describing some statistical technique that I was unfamiliar with. But, the more responses you write the more I'm convinced that you really don't get how basic statistical principles work.
No. You don't seem to grasp that statistical significance testing is a kind of probability analysis. The assumption initially made by Gosset was that "nature" randomly chooses an experimental outcome from a semi-infinite population of potental outcomes. If you know the probability distribution for the population you can determine the probability of various experimental outcomes. Gosset developed ways to estimate this population distribution from a small sample taken from the population. The larger the sample size, the higher the DoF and the better the estimate. For coin filps the population distribution of all outcomes is known exactly, and so DoF never enters into the probability calculation.

For large DoF the probability distribution becomes insensitive to DoF. With sufficiently large DoF you know the probability distribution of the population and it becomes like the coin example. You can do probability calculations.

Here's an example. You have a production line producing sequential batches. A long record of historical outcomes exists (DoF=1000). Analysis shows that the t-distribution fits the historical probability distribution very well. Based on this distribution you can draw your control chart showing expected outcome (mean) and the control lines placed at 1, 2, 3 standard deviations below and above the mean. Suppose you get a "4 sigma" outcome--one that is literally off the (control) chart. Is this significant? This sitation is analogous to the poker example above. If the extreme batch happens "out of the blue" then N is the number of batches that have been or might ever be produced, and so is very large. With large N the probability approaches 1, and the the chart-breaking batch is not likley to be due to something outside of chance. The null hypothesis is not rejected.

Suppose the process was perturbed in way that is expected (ie. predicted) to affect the process. The perturbation was stopped as soon as it was discovered. During the time when the perturbation was operating one batch was made. This batch was an extreme outlier. This situation is analogous to the predicted poker hand actually being dealt on the very next hand. N is then 1 and the probability that the bad batch reflected chance and not the perturbation is very small. The null hypothesis (that the perturbation has no effect) would be rejected.
Last edited by Mikebert; 12-26-2013 at 09:49 AM.







Post#4062 at 12-26-2013 10:47 AM by pbrower2a [at "Michigrim" joined May 2005 #posts 15,014]
---
12-26-2013, 10:47 AM #4062
Join Date
May 2005
Location
"Michigrim"
Posts
15,014

In any event, hurricanes, tornadoes, and now some winter storms are the extreme end of the distribution of storms. As an analogy in professional sports, hitters in baseball are not randomly distributed in talent. Major-league baseball has the top end of the distribution. At the very top -- there is only one Miguel Cabrera at the time. At the bottom in major-league baseball there is talent just barely in the majors -- what George Brett (the best hitter in his time, but by no means Miguel Cabrera) called the "Mendoza Line", the .200 hitter who barely stays on a major-league roster. The minor leagues are full of people who could hit .180 or so with no power if they made it to the major leagues; those people just don't make it to the Big Show. Most teams have some .220 hitters that they would replace in an instant, but as a rule it's easier to find someone who can hit .250 with slight power than to find someone who can hit .260 with enormous power -- and impossible to find a new Miguel Cabrera. Of course if one sees every team having people in AAA ball hitting .370 with extreme power (such people have short minor-league careers), then the game is changing somehow.

Hurricanes, tornadoes, and now some winter storms (the contrast between -10C air masses and juiced 25C air masses can force an extreme storm, and global warming makes that all the more likely -- witness Sandy) are at the extreme end of a continuum that begins with slight showers to Category 5 hurricanes and tornadoes. More frequent and more severe storms at the extreme end indicate that the game has changed, but the 'hitter' is changing the world in ways unwelcome in the extreme.
The greatest evil is not now done in those sordid "dens of crime" (or) even in concentration camps and labour camps. In those we see its final result. But it is conceived and ordered... in clean, carpeted, warmed and well-lighted offices, by (those) who do not need to raise their voices. Hence, naturally enough, my symbol for Hell is something like the bureaucracy of a police state or the office of a thoroughly nasty business concern."


― C.S. Lewis, The Screwtape Letters







Post#4063 at 12-26-2013 02:42 PM by The Grey Badger [at Albuquerque, NM joined Sep 2001 #posts 8,876]
---
12-26-2013, 02:42 PM #4063
Join Date
Sep 2001
Location
Albuquerque, NM
Posts
8,876

When one can predict a person's position on global warming simply by knowing what side of the political fence that person is on, then all these statistics and everything else are merely snow-coated mudballs in the endless partisan fight. I'm sorry. Knowing *how* your positions are determined, I find your scientific arguments to be specious.

That "you " is second person *plural*, BTW.
How to spot a shill, by John Michael Greer: "What you watch for is (a) a brand new commenter who (b) has nothing to say about the topic under discussion but (c) trots out a smoothly written opinion piece that (d) hits all the standard talking points currently being used by a specific political or corporate interest, while (e) avoiding any other points anyone else has made on that subject."

"If the shoe fits..." The Grey Badger.







Post#4064 at 12-26-2013 06:27 PM by pbrower2a [at "Michigrim" joined May 2005 #posts 15,014]
---
12-26-2013, 06:27 PM #4064
Join Date
May 2005
Location
"Michigrim"
Posts
15,014

I'll go with the statistical models, of course.
The greatest evil is not now done in those sordid "dens of crime" (or) even in concentration camps and labour camps. In those we see its final result. But it is conceived and ordered... in clean, carpeted, warmed and well-lighted offices, by (those) who do not need to raise their voices. Hence, naturally enough, my symbol for Hell is something like the bureaucracy of a police state or the office of a thoroughly nasty business concern."


― C.S. Lewis, The Screwtape Letters







Post#4065 at 12-26-2013 06:38 PM by Deb C [at joined Aug 2004 #posts 6,099]
---
12-26-2013, 06:38 PM #4065
Join Date
Aug 2004
Posts
6,099

You Think You Know How Much We Depend On Fossil Fuels, But You Have No Idea

This is really worth the two minutes it takes to get the story of fossil fuels and its impact on our world.

http://www.upworthy.com/you-think-yo...o-idea-2?c=to3
"The only Good America is a Just America." .... pbrower2a







Post#4066 at 12-26-2013 09:33 PM by Brian Beecher [at Downers Grove, IL joined Sep 2001 #posts 2,937]
---
12-26-2013, 09:33 PM #4066
Join Date
Sep 2001
Location
Downers Grove, IL
Posts
2,937

Quote Originally Posted by Deb C View Post
You Think You Know How Much We Depend On Fossil Fuels, But You Have No Idea

This is really worth the two minutes it takes to get the story of fossil fuels and its impact on our world.

http://www.upworthy.com/you-think-yo...o-idea-2?c=to3
I have yet to read your link but I will get to it probably tomorrow. In the meantime, not long ago it was said that for the young people coming of age today getting a driver's license and then a car is not the supreme rite of passage that it was for earlier generations. If that is the case, why are we today still just as dependent on the automobile for travel as we ever were, with no real effort at reducing said dependency? And do you think said reduction could happen withing our lifetimes?







Post#4067 at 12-27-2013 08:00 PM by Vandal-72 [at Idaho joined Jul 2012 #posts 1,101]
---
12-27-2013, 08:00 PM #4067
Join Date
Jul 2012
Location
Idaho
Posts
1,101

Quote Originally Posted by Mikebert View Post
And so can multiple events.
But multiple events allow you to compare distribution patterns, you can actually do a statistical analysis. A single event is not a pattern of anything.

What? Of course there is a difference.
No there isn't. You are assuming there is because to you that particular hand has meaning but to demonstrate that something was amiss you will need other information besides the result of the deal.

The probability of a being dealt a royal flush in N hands is given by

probability = 1 - (1-p)^N

Where p is 1/649740 and N is the number of hands being considered. When a person announces he is going to get a royal flush on the next deal he is saying it will occur in that hand and that hand only. N = 1 and the probability is p. Since p is very small it is very likley that a special cause is involved and I would strongly suspect the guy is cheating. I cannot believe you would hold no suspicions.
Do you have any actual evidence of cheating?

Now if someone gets dealt a royal flush out of the blue the exact same formula applies. But what is N? In this case no constraints that put an upper limit on N have been supplied (no predictions were made). So N is going to be the number of all poker hands that have been or will be dealt, because the set of hands under consideration was not specified. That is, N is going to be a really big number. For large N the expression above evaluates to 1. So the the probability is essentially 100%. In the first case, the null hypothesis is rejected, in the second it is not.
Probabilities of events and statistical significance of patterns are not the same thing at all.

A suspicion is not statistical evidence.

I read the article and I saw no mention of degress of freedom (DoF). Did you?
No. Because degrees of freedom were irrelevant to that particular example because we weren't actually doing a statistical test! It was a way to calculate a minimum sample size necessary. It is telling you what the degrees of freedom will be if you properly run such a test! You really don't understand this as well as you think.

And this is the issue. You apply formulas. You do not understand what you are doing when you apply these formulas. So you are making an argument from authority.
You don't actually have a real statistical test for significance. Got it. You are just making stuff up. Got it.

Maximum sustained wind speed is not a continuous variable: there is one value per storm.
That is most certainly a continuous variable! 123 mph, 67.3 mph, 112.654 mph. If the number of possible values is dependent on your level of precision, then the variable is continuous. Simple statistics man.

It is explicitly used to categorize storms.
That is NOT what you originally claimed you were going to do with your four sigma storm! How exactly would you calculate a four sigma category of a storm? There are many statistical tests that can be run on categorical data and there are statistical tests for continuous data. You started out making a claim about continuous data (a four sigma wind speed storm). I took you at your word and pointed out that the data is continuous, not discrete. Now you are bringing up categorical data despite it having nothing to do with your original claim. Trust me, I'm not the confused one.

No. You don't seem to grasp that statistical significance testing is a kind of probability analysis. The assumption initially made by Gosset was that "nature" randomly chooses an experimental outcome from a semi-infinite population of potental outcomes. If you know the probability distribution for the population you can determine the probability of various experimental outcomes. Gosset developed ways to estimate this population distribution from a small sample taken from the population. The larger the sample size, the higher the DoF and the better the estimate. For coin filps the population distribution of all outcomes is known exactly, and so DoF never enters into the probability calculation.
So, how do you determine if a particular coin is biased? Is there a minimum number of flips necessary?

Is there a minimum number of card deals necessary to determine if a dealer is biased?

Is is there a minimum number of storm measurements necessary to determine if the storm source is biased?

For large DoF the probability distribution becomes insensitive to DoF. With sufficiently large DoF you know the probability distribution of the population and it becomes like the coin example. You can do probability calculations.
Do you really not see why this completely demolishes your original claim?

Here's an example. You have a production line producing sequential batches. A long record of historical outcomes exists (DoF=1000). Analysis shows that the t-distribution fits the historical probability distribution very well. Based on this distribution you can draw your control chart showing expected outcome (mean) and the control lines placed at 1, 2, 3 standard deviations below and above the mean. Suppose you get a "4 sigma" outcome--one that is literally off the (control) chart. Is this significant?
No. You have changed the meaning of the word "significant". You are using the lay word "important" instead of the proper statistical meaning of the word significant.

In your example do you conclude that the entire production line is faulty and must be torn apart and serviced at great expense from a single anomalous result? Or, do you wait to see if the four sigma event or multiple three sigma events are produced often enough to conclude that the original distribution pattern is no longer valid? If you choose the first option you will be costing your company huge amounts of money in wasted time and materials because you don't grasp what four sigma really means.

This sitation is analogous to the poker example above. If the extreme batch happens "out of the blue" then N is the number of batches that have been or might ever be produced, and so is very large. With large N the probability approaches 1, and the the chart-breaking batch is not likley to be due to something outside of chance. The null hypothesis is not rejected.
Once again, probability of an event and statistical significance are not the same thing.

Suppose the process was perturbed in way that is expected (ie. predicted) to affect the process.
You can only predict this because you have outside information. This is no longer analogous to your storm claim. We don't have alternate Earth's. We don't understand all the physics involved in storm production like we do in a production line. This argument is beginning to sound like a corollary to the Salem Hypothesis.

The perturbation was stopped as soon as it was discovered. During the time when the perturbation was operating one batch was made. This batch was an extreme outlier. This situation is analogous to the predicted poker hand actually being dealt on the very next hand. N is then 1 and the probability that the bad batch reflected chance and not the perturbation is very small. The null hypothesis (that the perturbation has no effect) would be rejected.
Your analogy does not match the original claim. We don't know that increased warming will generate larger storms, we only suspect it will. We run the analysis to determine if our suspicions are accurate. Waiting for a single large storm and declaring the analysis is over is exactly the pseudo-scientific crap that parapsychology kooks pull. It is statistical fraud!







Post#4068 at 12-28-2013 11:07 AM by B Butler [at joined Nov 2011 #posts 2,329]
---
12-28-2013, 11:07 AM #4068
Join Date
Nov 2011
Posts
2,329

Left Arrow Time to Crunch Numbers?

Quote Originally Posted by Vandal-72 View Post
In your example do you conclude that the entire production line is faulty and must be torn apart and serviced at great expense from a single anomalous result? Or, do you wait to see if the four sigma event or multiple three sigma events are produced often enough to conclude that the original distribution pattern is no longer valid? If you choose the first option you will be costing your company huge amounts of money in wasted time and materials because you don't grasp what four sigma really means.
Mike

Suppose we have a data base of hundreds to thousands of (tropical?) storms with a reasonable estimate of max wind speed for each storm. Suppose we cull this data base to include only samples that occurred "before global warming really took off". Could anyone guess/calculate the max wind speed a single modern storm would need to get tagged 'four sigma' against such a distribution?

I'm beginning to suspect the "Winston" effect. People who live in ivory towers might not have a gut instinctive feel for just how unusual a real world four sigma event is. Real numbers might or might not help.
Last edited by B Butler; 12-28-2013 at 09:39 PM. Reason: Added a link







Post#4069 at 12-28-2013 12:25 PM by Deb C [at joined Aug 2004 #posts 6,099]
---
12-28-2013, 12:25 PM #4069
Join Date
Aug 2004
Posts
6,099

The following piece gives me hope because people are taking the lead in regards to addressing major issues, such as Global warming and our economic problems. As the article indicates, it is usually pop culture that explores change before politicians do.

10 Hopeful Things That Happened in 2013 to Get You Inspired for What’s to Come

Beyond the headlines of conflict and catastrophe, this year’s top stories offered us some powerful proof that the world can still change—for the better.

Much happened that was hopeful this year—a new pope focused on inequality, successful minimum wage campaigns spread across the country, and the number of states allowing gay marriage doubled.But responses to the threat of the climate crisis lead off this year’s top stories as we look at seeds sown this year that could make 2014 transformational.

1. We saw surprising new leadership on the climate issue

In northeast Nebraska, Native Americans and local ranchers formed a new allianceto resist the Keystone XL pipeline. Seven thousand activists gathered in Pittsburghto press for action on a wide range of environmental justice issues. Students across North America persuaded nine colleges and universities to divest from fossil fuel companies. Hundreds of climate activists walked out of the COP19 climate talks in Poland to hold their own climate talks.

The governors of California, Oregon, Washington, and the Canadian province of British Columbia have committed to taking action on the climate crisis. But Congress remains deadlocked and in denial, and climate scientists—when they let down their careful professional demeanor—express astonishment that world governments have failed to act on what is fast becoming a global emergency.

A new potential ally is coming from an unexpected source. Some investors are beginning to worry that fossil fuel companies may not be a good bet. Investors worry about a “carbon bubble.”
The reserves of oil, gas, and coal counted as assets by the big energy corporations would be enormously destructive to life on Earth if they were allowed to burn. Many believe that new regulation or pricing will keep a large portion of those reserves safely in the ground.

If that happens, the companies' reserves, and thus their stock, may be worth far less than believed. Savvy investors are placing their bets elsewhere: Warren Buffett, for example, is investing $1 billion in wind energy, which, along with solar energy, is looking better all the time.
Included in this piece are two other examples of how average citizens have taken action.

2. Native peoples took the lead in the fossil fuel fight


3. The middle and lower classes fought for economic justice


http://www.yesmagazine.org/people-po...paign=20131227
"The only Good America is a Just America." .... pbrower2a







Post#4070 at 12-29-2013 12:21 PM by B Butler [at joined Nov 2011 #posts 2,329]
---
12-29-2013, 12:21 PM #4070
Join Date
Nov 2011
Posts
2,329

Left Arrow AGU talk on science and advocacy

RealClimate is featuring a talk on science and advocacy.

We have often discussed issues related to science communication on this site, and the comment threads frequently return to the issue of advocacy, the role of scientists and the notion of responsibility. Some videos from the recent AGU meeting are starting to be uploaded to the AGU Youtube channel and, oddly, the first video of a talk is my Stephen Schneider lecture on what climate scientists should advocate for (though actually, it mostly about how science communicators should think about advocacy in general since the principles are applicable regardless of the subject area):

The talk has provoked a number of commentaries from the Yale forum, Andy Revkin at DotEarth,Judith Curry and Bart Verhegggen – with varying degrees of comprehension of the main points.


Just thought I'd throw it up.







Post#4071 at 12-30-2013 02:54 PM by Mikebert [at Kalamazoo MI joined Jul 2001 #posts 4,502]
---
12-30-2013, 02:54 PM #4071
Join Date
Jul 2001
Location
Kalamazoo MI
Posts
4,502

So, how do you determine if a particular coin is biased? Is there a minimum number of flips necessary?
Yes. Using the reference you cited it gives an example of calculating the number of flips necessary to establish that a coin is fair (has a true heads probability p of 0.5) with less than 0.01 error and a statistical significance of 95% (Z = 1.96). The result is 9604. This analysis shows it takes nearly 10000 flips to determine that p is between 0.049 and 0.51. For a biased coin with an assumed p of 0.65 it takes 8740 flips to determine that p is between 0.64 and 0.66. But we are interested in whether or not the coin is biased, not in exactly how biased it is. With 350 flips we can determine that p is between 0.60 and 0.70 (E = 0.05) with 95% confidence (Z = 1.96). We don't need the 0.01 precision for this purpose.

As you increase p is the number of flips does down. Suppose we are testing a coin with an assumed bias of 0.75? It would take 32 flips to determine that p was between 0.6 and 0.9 (E = 0.15). With p = 0.8 we can establish that p is between 0.6 and 1 with 16 rolls. With p = 0.9 we can establish that p is between 0.6 and 1.0 at 95% confidence with just 4 rolls and so on. What this means is if more than one tails is often, the hypothesis of extreme bias (p = 0.9) is rejected.

In the case of p =1 (e.g. a 2 headed coin) rolling a single tails is definitive. A coin that rolls a tails cannot be a 100% head coin. Suppose you are asked to test whether or not a given coin was the all-heads coin. You flip the coin and it comes up tails. It cannot be the all-head coin. Do you see this?

Suppose there is a virtually all-heads coin, that is, p is different from 1 by an infinitesimal e. You are asked to test whether this coin was the virtually all-heads coin or not with a 95% confidence. You flip this coin and it comes up tails, an outcome with infinitesimal probability e. You can say that this coin is not a virtually all-heads coin with a high degree of confidence (higher than 95% for sufficienty small e). The smaller the value of e, the higher the degree of confidence. As e approaches zero, the degree of confidence approaches certainty as we see with the two-headed coin. For sufficiently small e only a single roll is needed to decide the issue for a given degree of confidence. This is why the number of rolls in the formulas from the reference you cited decreases as p approaches 1. There are cases in which a single flip is enough.

You can only predict this because you have outside information. This is no longer analogous to your storm claim.
I constructed it to be analogous to the poker example. It addresses your claim that there is no difference between a player getting dealt a royal flush out of the blue and being dealt one after making a prediction that he would do so. The prediction is the outside information.

Your analogy does not match the original claim. We don't know that increased warming will generate larger storms, we only suspect it will.
And we don’t know the guy claiming he will get a royal flush is not full of shit. But if he doesn’t get a royal flush then we can conclude that he is full of shit. And we don’t know that the highly biased coin is actually highly biased, which is why we are testing it. And if it rolls a tail right off then we can conclude with a high degree of confidence that the claim that it (virtually) always rolls heads is bogus.







Post#4072 at 12-30-2013 03:37 PM by Vandal-72 [at Idaho joined Jul 2012 #posts 1,101]
---
12-30-2013, 03:37 PM #4072
Join Date
Jul 2012
Location
Idaho
Posts
1,101

Quote Originally Posted by Mikebert View Post
Yes. Using the reference you cited it gives an example of calculating the number of flips necessary to establish that a coin is fair (has a true heads probability p of 0.5) with less than 0.01 error and a statistical significance of 95% (Z = 1.96). The result is 9604. This analysis shows it takes nearly 10000 flips to determine that p is between 0.049 and 0.51. For a biased coin with an assumed p of 0.65 it takes 8740 flips to determine that p is between 0.64 and 0.66. But we are interested in whether or not the coin is biased, not in exactly how biased it is. With 350 flips we can determine that p is between 0.60 and 0.70 (E = 0.05) with 95% confidence (Z = 1.96). We don't need the 0.01 precision for this purpose.

As you increase p is the number of flips does down. Suppose we are testing a coin with an assumed bias of 0.75? It would take 32 flips to determine that p was between 0.6 and 0.9 (E = 0.15). With p = 0.8 we can establish that p is between 0.6 and 1 with 16 rolls. With p = 0.9 we can establish that p is between 0.6 and 1.0 at 95% confidence with just 4 rolls and so on. What this means is if more than one tails is often, the hypothesis of extreme bias (p = 0.9) is rejected.

In the case of p =1 (e.g. a 2 headed coin) rolling a single tails is definitive. A coin that rolls a tails cannot be a 100% head coin. Suppose you are asked to test whether or not a given coin was the all-heads coin. You flip the coin and it comes up tails. It cannot be the all-head coin. Do you see this?

Suppose there is a virtually all-heads coin, that is, p is different from 1 by an infinitesimal e. You are asked to test whether this coin was the virtually all-heads coin or not with a 95% confidence. You flip this coin and it comes up tails, an outcome with infinitesimal probability e. You can say that this coin is not a virtually all-heads coin with a high degree of confidence (higher than 95% for sufficienty small e). The smaller the value of e, the higher the degree of confidence. As e approaches zero, the degree of confidence approaches certainty as we see with the two-headed coin. For sufficiently small e only a single roll is needed to decide the issue for a given degree of confidence. This is why the number of rolls in the formulas from the reference you cited decreases as p approaches 1. There are cases in which a single flip is enough.
Is that at all relevant to your storm example?

I constructed it to be analogous to the poker example. It addresses your claim that there is no difference between a player getting dealt a royal flush out of the blue and being dealt one after making a prediction that he would do so. The prediction is the outside information.
No, it isn't. The prediction is not evidence of cheating. You can not use something that catches your initial attention as supporting proof of your explanation for the event.

Your four sigma storm can not be both an indication that something is different and proof for why that difference exists. Statistical fraud again.

And we don’t know the guy claiming he will get a royal flush is not full of shit. But if he doesn’t get a royal flush then we can conclude that he is full of shit.
Try this: you state your intention of getting the royal flush and you know you are not colluding with the dealer. Do you conclude that the dealer is crooked if you do get the royal flush?

And we don’t know that the highly biased coin is actually highly biased, which is why we are testing it. And if it rolls a tail right off then we can conclude with a high degree of confidence that the claim that it (virtually) always rolls heads is bogus.
Is your "always heads" prediction really analogous to your storm example since coin results are categorical data and wind speeds are continuous?







Post#4073 at 12-31-2013 08:28 AM by Mikebert [at Kalamazoo MI joined Jul 2001 #posts 4,502]
---
12-31-2013, 08:28 AM #4073
Join Date
Jul 2001
Location
Kalamazoo MI
Posts
4,502

Quote Originally Posted by Vandal-72 View Post
Your four sigma storm can not be both an indication that something is different and proof for why that difference exists.
Supporting proof is provided by the result of the experiment or study.

Statistical significance testing is used to show indication that something is different. That is, is the result different from what would happen without the treatment?

Suppose one comes up a process modification shown that increases yield in the laboratory. It is tried in a 15 lot production campaign and the average yield was 83.2% with standard deviation 2.3%. The average of the previous 600 lots run with the SOP process yielded 85.1% with standard deviation 2.4%. At the end of 15 lot test the process engineer concluded that the new process did not increase yield and goes back to the SOP. Is there any need for statistical analysis? That is, will it change the conclusion that the change did not increase yield?

Suppose the yield had been 84.7%. Does this change the conclusion that the change did not increase yield?
Last edited by Mikebert; 12-31-2013 at 08:50 AM.







Post#4074 at 12-31-2013 08:45 AM by pbrower2a [at "Michigrim" joined May 2005 #posts 15,014]
---
12-31-2013, 08:45 AM #4074
Join Date
May 2005
Location
"Michigrim"
Posts
15,014

Quote Originally Posted by Mikebert View Post
Supporting proof is provided by the result of the experiment or study.

Statistical significance testing is used to show indication that something is different. That is, is the result different from what would happen without the treatment?

Suppose one comes up a process modification shown that increases yield in the laboratory. It is tried in a 15 lot production campaign and the average yield was 84.7% with standard deviation 2.3%. The average of the previous 600 lots run with the SOP process yielded 85.1% with standard deviation 2.4%. At the end of 15 lot test the process engineer goes back to the SOP, concluding that there is no reason to keep the change. Is there any need for statistical analysis? That is, will it change the conclusions? Why or why not?
Quote Originally Posted by Wikipedia
A type I error (or error of the first kind) is the incorrect rejection of a true null hypothesis. It is a false positive. Usually a type I error leads one to conclude that a supposed effect or relationship exists when in fact it doesn't. Examples of type I errors include a test that shows a patient to have a disease when in fact the patient does not have the disease, or an experiment indicating that a medical treatment should cure a disease when in fact it does not.

A type II error (or error of the second kind) is the failure to reject a false null hypothesis. It is a false negative. Examples of type II errors would be a blood test failing to detect the disease it was designed to detect, in a patient who really has the disease; or a clinical trial of a medical treatment failing to show that the treatment works when really it does.
https://en.wikipedia.org/wiki/Type_I_and_type_II_errors

In that case there is little to lose by taking the chance and finding no difference. That of course assumes that the consequences for making a wrong choice are trivial. In medical testing the consequences of mistakenly believing that one has a better treatment when one has none are severe and have irretrievable consequences. For a production line accepting a Type I error is to get an illusory gain. So one changes something and gets no result. Failure to make the effort means that one surrenders the opportunity to make a profitable change.
The greatest evil is not now done in those sordid "dens of crime" (or) even in concentration camps and labour camps. In those we see its final result. But it is conceived and ordered... in clean, carpeted, warmed and well-lighted offices, by (those) who do not need to raise their voices. Hence, naturally enough, my symbol for Hell is something like the bureaucracy of a police state or the office of a thoroughly nasty business concern."


― C.S. Lewis, The Screwtape Letters







Post#4075 at 12-31-2013 09:12 AM by Mikebert [at Kalamazoo MI joined Jul 2001 #posts 4,502]
---
12-31-2013, 09:12 AM #4075
Join Date
Jul 2001
Location
Kalamazoo MI
Posts
4,502

Quote Originally Posted by pbrower2a View Post
https://en.wikipedia.org/wiki/Type_I_and_type_II_errors

In that case there is little to lose by taking the chance and finding no difference. That of course assumes that the consequences for making a wrong choice are trivial. In medical testing the consequences of mistakenly believing that one has a better treatment when one has none are severe and have irretrievable consequences. For a production line accepting a Type I error is to get an illusory gain. So one changes something and gets no result. Failure to make the effort means that one surrenders the opportunity to make a profitable change.
I didn't see where this applies. The result was that the change made yields lower. No statistical analysis was done so there is no chance to have type I or type II errors. The question asked was would statistical analysis change the conclusion that the process change does not increase yield?
-----------------------------------------