Climate Change. Statistics 101 says its BS.

preaching to the choir.

 

Of course the "man-made global warming" science is bunk.

Our planet is currently in what is known as an "interglacial period" - which of course are warmer than "glacial periods" which are also known as "ice ages".

Mankind plays no role in these natural planetary temperature cycles.

 

It's a very good question. I raise the "statistics" argument every single time someone tries to tell me that "scientists" have proven global warming is real.

So far I haven't had anyone come close to refuting that fine little tidbit.

 

i think its extremely egotistical for man to think they can have such a huge impact on the planet in such a short period of time. im not talking deforestation or smog, but a global effect on how the earth works. crazy.

 

Scientist and "ego" pretty much go hand in hand. Walk into any college science class and you'll see what I mean. I love science, but you have to keep in mind that scientists are the ones that told us the earth was flat and the center of the universe.

It's amazing how far we've come yet nothing is truly any different. We still think we have it all figured out.

 

Preface: I do not believe in global warming. I do believe that we as a world need to clean up our act, specifically china. There are proven effects of pollution.

The problem, we will never have 4.7 billion years of data. So scientist have to take what information is available -200 years. Their studies are just trying to show the record for that recorded time. In some of their OPINIONS it means climate is changing more drastically. They are called theories for a reason and you can't not try/study them just because there is not enough information. Scientist have to make educational guesses.

 

That's not true at all. As long as it's a random sample you can certainly draw conclusions from a sample that size; you will just have large statistical uncertainty.

"Weather" hasn't been around for 4.7 billion years, first of all. Secondly, it depends on what one is trying to do. If you're trying to establish what the average temperature for a particular date in a particular region is, for example, would you argue that you need to know all 4.7 billion values of that number to calculate the mean? Of course not.

Trying to extrapolate what will happen in the long term from a data sample that only goes back 150 years is, of course, problematic and highly dependent on the particular model one uses.

That is absolutely not a true statement. There certainly is enough data to draw conclusions and establish confidence limits on those conclusions.

HOWEVER, that's not the same thing as saying that the climate change people are right. I'm just saying there is a lot of valid science one can do from a dataset that size.

EDITED TO ADD: And of course, as a later poster pointed out, there is a much larger data set than that. Isotope data goes back to 100 Mya+

 

Not really. "Science" as we know it didn't develop as a discipline until much later. The flat earth and the center of the universe was given to us by "philosophers," which is not at all the same thing.

By the time science rolled around, it was the Church who insisted those things were true, and burned people at the stake for suggesting that the Earth might revolve around the Sun.

No real scientist thinks we have anything figured out.

 

Do you understand statistics? Sampling errors and confidence limits?

I'm not trying to challenge you or anything, I am genuinely curious. I find that few people really do.

 

Been years since I have done any statistical calculations - and I am too lazy to mess with it now-

But when ever you see statistics reported -

Like 45% approve 55% disapprove

The margin of error should also be presented -

Normal political poll will be 1,000 people sample with a margin of error of 3% (whatever)

Which of course means that the actual numbers could be

42-48% approve 52-58% disapprove

If you see statistics presented and they do not give the margin of error - or in some cases it may be presented as confidence interval - then they have not really given you any usable information.

It is possible to use very small samples and still have valid results -

Another way to look at it -(I am GUESSING) --

If you increase the sample size from 1,000 to 2,000 your margin of error may drop from 3% down to 2.995%

So the statistician must ask - is the extra effort of making the sample size larger worth it - does it reduce the margin of error enough to justify the extra expense?

The surprise is how little impact increasing the sample size has on reducing the margin of error.

 

The error on the mean decreases as 1/(sqrt(n)) for a random sample. So going from 1k to 2k gives you a factor of 1 over root 2 improvement assuming your sampling techinique is truly random. Expressed in a dimensionless parameter like percent, doubling your sample size takes you from 3% to ~2.12%

 

There is MUCH more data than just the last 150 years:

 

It pollutes OUR air.
http://geology.com/nasa/monitoring-pollution-by-satellite.shtml

http://onigirisensei.wordpress.com/2008/04/05/china-most-polluted-country/

 

Back in the 70's wasn't the mantra "The ice age is coming by the year 200" in the 80's and 90's it was " We are all going to burn because of Global warming" Now its Climate Change, hotdog finally something that we can all agree on, the climate does change. Simple as that.DUH Yesterday it rained and today its warm and dry, been that way for millions of years, heck who knows it may snow tomorrow....

 

I trust your calculations - like I said its been years -

I do recall going through some analysis - where you increase the sample size to reduce the margin of error to an acceptable level - is the cost of a greater sample size worth the extra $ to reduce your margin of error.

So if going from 1K to 2K sample takes you from 3% to 2.12% margin of error

What sample size would it take to get the margin of error down to 1%? or .5%?

I seem to recall getting to a point where increasing the sample size by a lot only decreased the margin of error by a little.