Debates
16 Apr 05
16 Apr 05
1 edit
Ivanhoe wrote:
"Right, correlations ..... the statistical way of thinking is the road to nowhere if you want to analyse things."
I notice a trend in science to use statistical inference (really an exploratory tool or a weapon of last resort) in situations where it might be possible to create some theoretical model based on data and general principles and apply more analytical scientific reasoning. Then again, there are strong correlations (😉) between inferential statistics and the hypothesis-testing of experimental science. What is the best approach (in a given situation)?
16 Apr 05
4 edits
Originally posted by royalchickenhttp://nkmr.org/english/a_jury_gives_its_verdict_on_meadows_law.htm
Ivanhoe wrote:
"Right, correlations ..... the statistical way of thinking is the road to nowhere if you want to analyse things."
I notice a trend in science to use statistical inference (really an exploratory tool or a weapon o ...[text shortened]... mental science. What is the best approach (in a given situation)?
A jury gives its verdict on Meadows Law
By Eric Roberts, reporter
"Sir Roy's view that one sudden infant death is a tragedy, two is suspicious and three is murder has become known as Meadow's Law, and his evidence before the Clark case had always been accepted virtually without question by police, prosecutors and courts."
The above article shows that "statistical" thinking sends innocent people to jail .... The Attorney-General in England is reviewing more than 250 cases where a parent may have been wrongly convicted ......
or: "How Highly Intelligent People Can Be Stupid As A Mule's Ass."
http://nkmr.org/english/a_jury_gives_its_verdict_on_meadows_law.htm
16 Apr 05
Originally posted by ivanhoeThanks for the link. My stats lecturer actually discussed how the problem in this case was not the result of statistical thinking but of invalid statistical thinking. I'm not knocking statistics as a science (some of its results are shockingly useful), but its use when more conventional reasoning is possible.
http://nkmr.org/english/a_jury_gives_its_verdict_on_meadows_law.htm
A jury gives its verdict on Meadows Law
By Eric Roberts, reporter
"Sir Roy's view that one sudden infant death is a tragedy, two is suspicious and three is murder has become known as Meadow's Law, and his evidence before the Clark case had always been accepted virtually witho ...[text shortened]... As A Mule's Ass."
http://nkmr.org/english/a_jury_gives_its_verdict_on_meadows_law.htm
17 Apr 05
Originally posted by royalchicken
Thanks for the link. My stats lecturer actually discussed how the problem in this case was not the result of statistical thinking but of invalid statistical thinking. I'm not knocking statistics as a science (some of its results are shockingly useful), but its use when more conventional reasoning is possible.
Of course it is invalid statistical thinking. I find it unbelievable that this could go on and on for such a long time without somebody intervening ..... mindboggling.
22 Apr 05
Originally posted by royalchickenIf we ever get close to joining the Quantum world with the Relative world, what part do you think Statistical Analysis will play?
Thanks for the link. My stats lecturer actually discussed how the problem in this case was not the result of statistical thinking but of invalid statistical thinking. I'm not knocking statistics as a science (some of its results are shockingly useful), but its use when more conventional reasoning is possible.
I have thought about this a lot. Especially if "strings" turn out to be more than beauty.
The only tool we have left on "things" that can't be measured or 'broken' seems to be analysis. But That's why i'll probably be happy to just continue on the road to becoming a worn out VB6 kind of guy.
22 Apr 05
Originally posted by DoctorScribblesThat's one data point. Don't we need a few more to see if it belongs to a statistical trend? Perhaps this point falls outside the standard deviation? Or more likely, it is the result of a mechanical error and we can disregard this point all together.
I'm sorry to report that this is nonsense.
22 Apr 05
Originally posted by ColettiYour point is well taken.
That's one data point. Don't we need a few more to see if it belongs to a statistical trend? Perhaps this point falls outside the standard deviation? Or more likely, it is the result of a mechanical error and we can disregard this point all together.
However, if ivanhoe had claimed "the statistical way of analyzing other people's opinions is the road to nowhere if you want to analyse that about which they opine," I would not have taken issue with it.
22 Apr 05
Originally posted by royalchickenCan you offer an example?
Ivanhoe wrote:
"Right, correlations ..... the statistical way of thinking is the road to nowhere if you want to analyse things."
I notice a trend in science to use statistical inference (really an exploratory tool or a weapon of last resort) in situations where it might be possible to create some theoretical model based on data and general princ ...[text shortened]... e hypothesis-testing of experimental science. What is the best approach (in a given situation)?
22 Apr 05
1 edit
Originally posted by StarValleyWyI'm starting to get scared now. So far Coletti and SVW both appear to think that statistical analyisis about some subject of study essentially takes the form of polling people's opinions about that subject. Do others share this same warped view of what statistics is all about?
Sure. For every ten people who you ask a question... one will answer that it is ok to kill babies. A thousandyoung respondants will give you the correct analysis.
22 Apr 05
Originally posted by DoctorScribblesI surrender. You got me good. Sorry. <grin>
I'm starting to get scared now. So far Coletti and SVW both appear to think that statistical analyisis about some subject of study essentially takes the form of polling people's opinions about that subject. Do others share this same warped view of what statistics is all about?