CKA Forums
Login 
canadian forums
bottom
 
 
Canadian Forums

Author Topic Options
Offline
CKA Uber
CKA Uber
 Vancouver Canucks
User avatar
Profile
Posts: 20991
PostPosted: Wed Dec 11, 2013 9:38 am
 


andyt wrote:
In the end it's all what we believe. Believe it and you will see it.


What we call confirmation bias in science.


Offline
CKA Moderator
CKA Moderator
User avatar
Profile
Posts: 33120
PostPosted: Wed Dec 11, 2013 9:51 am
 


Zipperfish wrote:

There's a rampant confimration bias where scientists are rejecting the null hypothesis in order to find effects that aren't really there.

For non-science types, when you run a science you don't actually test your hypothesis you test the null hypothesis:

Hypothesis: Sugar makes kids hyperactive
Null Hypothesis: Sugar has no effect on children's hyperactivity.

In order to find an effect, you have to "reject the null hypothesis" (that sugar has no effect). Accepting the null hypothesis means you didn't find anything, which isn't going to get you published anywhere.


Well said.

Zipperfish wrote:
The use of statistics is also a problem. Uncertainties are not being accurately communicated. When you undergo a contrived statistical exercise in order to shopw the effect you're looking for, you tend to multiply existing uncertainties, until the uncertainty itself is greater than the effect you're looking for. If uncertainties are greater than the effect, you can't reject the null hypothesis, and you're not going to get published.


Just to add - Statistical analysis should always include the error factor. I took too many courses that failed you if that calculation wasn't there.

My favourite real world statistics are done by pollsters using a horribly small data set - like 120 people - and try to apply those conclusions to a population of 40 million. They never include the error rate, because it would be like "+/- 30% 3 times in 20" instead of a usual "+/- 5% 19 times out of 20".

And then people believe what the pollsters say, not realizing they aren't saying anything but instead relaying their own bias or that of whoever paid for the 'study' for you to adopt.


Offline
CKA Uber
CKA Uber
User avatar
Profile
Posts: 33600
PostPosted: Wed Dec 11, 2013 9:58 am
 


Zipperfish wrote:
andyt wrote:
In the end it's all what we believe. Believe it and you will see it.


What we call confirmation bias in science.


Yes, and I think it's rampant. Especially where money is involved. And as was pointed out, often it's just pure cheating.

Ultimately tho, I still think there is no truth "out there." I think we have deep biases built in, limits to what we can sense and comprehend. Different topic tho, I guess.


Offline
CKA Super Elite
CKA Super Elite
User avatar
Profile
Posts: 8280
PostPosted: Wed Dec 11, 2013 10:00 am
 


DrCaleb wrote:
andyt wrote:
In the end it's all what we believe. Believe it and you will see it. Seems to be most prevalent with medical research, because of the enormous bucks involved.


That's actually the exact opposite of Science. Belief does not enter into it. That's the trap many fall into, they can't believe it so it's not real.

Data is the key to good science. Reproducible results that confirm or deny a conclusion. Anyone can drop a feather and a rock off the leaning tower of Pisa and anyone can see and measure the results. If the feather hits the ground first one time, it tells us much more than the rock hitting first one million times.

Another big, especially in the social "sciences", problem is poor experimental design. Reminds me of a joke about the psychologist who was testing the effect of a vitamin on rats. He found 33.3% turn right when entering a maze, 33.3% turn left when entering the a maze and the third rat just stood there.


Offline
CKA Uber
CKA Uber
User avatar
Profile
Posts: 33600
PostPosted: Wed Dec 11, 2013 10:09 am
 


That was my point - we still believe a lot of these poorly done experiments, in part because nobody else bothers to try to replicate them. Especially in medicine, as has been pointed out. There's a strong incentive to find for a particular drug, say. Nobody else can replicate because the drug is patented, or if they could, it would cost a fortune and not be of any benefit for them. We're relying on the people who stand to profit to tell us what's what. Not a good system.


Offline
CKA Uber
CKA Uber
 Vancouver Canucks
User avatar
Profile
Posts: 20991
PostPosted: Wed Dec 11, 2013 10:15 am
 


andyt wrote:
Zipperfish wrote:
andyt wrote:
In the end it's all what we believe. Believe it and you will see it.


What we call confirmation bias in science.


Yes, and I think it's rampant. Especially where money is involved. And as was pointed out, often it's just pure cheating.

Ultimately tho, I still think there is no truth "out there." I think we have deep biases built in, limits to what we can sense and comprehend. Different topic tho, I guess.


There's epistemological limits in science, sure. Logical induction, for one, is on pretty shaky philosophical ground. Even logical deduction is tutrtles all the way down. Plus we are bouncing up against the limits of truth with Hesienberg's UNcertainty Prinnciple, Godel's Incompleteness Theorem and some of the stranger aspects of quantum mechanics. it's a fascinating field and I'll fill pages if we ever get a topci going on that.

But this is just plain wrong science. A good scientist should do his utmost to detsroy his own hypothesis, on the assumption that if she doesn't, someone else will do it for her, except publicly.





PostPosted: Wed Dec 11, 2013 10:24 am
 


BartSimpson wrote:
sandorski wrote:
BartSimpson wrote:

+5 R=UP


-5 PDT_Armataz_01_33


You're a dick.



+1000


From reading the purple hippos posts I get the feeling he is one of those "paid" to do bullshit science studies.


Offline
CKA Moderator
CKA Moderator
User avatar
Profile
Posts: 33120
PostPosted: Wed Dec 11, 2013 10:37 am
 


Zipperfish wrote:
Even logical deduction is tutrtles all the way down. Plus we are bouncing up against the limits of truth with Hesienberg's UNcertainty Prinnciple, Godel's Incompleteness Theorem and some of the stranger aspects of quantum mechanics. it's a fascinating field and I'll fill pages if we ever get a topci going on that.


I have just such an example I read lately!

Science has been searching for 'dark matter' for some time, with disappointing results. One bright man thought of a new way of detecting a particle know as a axion, a particle thought to have a mass so small it's the fraction of an Electron and it doesn't readily interact with other particles so it's difficult to detect.

This guy thought to use a microscopic detector called a "Pi Josephson junction" which should give off little bits of electricity whenever an Axion passes through it (interacting with neutrons slightly), which should happen fairly often. Much more often than the WIMPs that the dark matter experiments deep underground with billions of litres of water were designed to find.

Turns out, an experiment back in 2004 found exactly the electrical blips they were expecting to come from Axions, but the experimenters at the time dismissed the unknown blips as errors in the measurement equipment. The number of Axions, despite them being very light, could make up the mass we expect to find for 'dark matter' in the universe.


Offline
CKA Uber
CKA Uber


GROUP_AVATAR
User avatar
Profile
Posts: 20852
PostPosted: Wed Dec 11, 2013 10:53 am
 


DrCaleb wrote:
It used to be that Universities did research for research's sake alone. Now, they make the number of published articles as a measure for the value of a researcher.

Professor Higgs only published 5 papers in his career, and won a Nobel Prize. He was recently quoted as saying that because of this new 'Publish! Publish! Publish!' mentality, he'd never have kept his job in today's work environment.

But the point this guy I think is making is that science belongs to everyone, not just people who subscribe to a certain journal; so publish it where it's accessible to everyone freely.


Well, it used to be that a lot of research at universities was largely funded by governments (except perhaps the USA).

In the past generation or so, they've scaled back that funding in a big way and someone had to step in so that they could keep running (especially here in Canada where most universities are limited to how much tuition they can charge).

Who had millions of dollars to spend on research? Why corporations of course!

That led to a boom in research to access those funds and likely was one of the reason the 'Publish or Perish' phenomenon came onto the scene. I remember one of my professors being ecstatic one day and telling me "I have been quoted and therefore I exist!" I doubt that happened 50 or 60 years ago.

Excellent read BTW! R=UP


Offline
CKA Uber
CKA Uber
 Vancouver Canucks
User avatar
Profile
Posts: 20991
PostPosted: Wed Dec 11, 2013 12:34 pm
 


DrCaleb wrote:
Zipperfish wrote:
Even logical deduction is tutrtles all the way down. Plus we are bouncing up against the limits of truth with Hesienberg's UNcertainty Prinnciple, Godel's Incompleteness Theorem and some of the stranger aspects of quantum mechanics. it's a fascinating field and I'll fill pages if we ever get a topci going on that.


I have just such an example I read lately!

Science has been searching for 'dark matter' for some time, with disappointing results. One bright man thought of a new way of detecting a particle know as a axion, a particle thought to have a mass so small it's the fraction of an Electron and it doesn't readily interact with other particles so it's difficult to detect.

This guy thought to use a microscopic detector called a "Pi Josephson junction" which should give off little bits of electricity whenever an Axion passes through it (interacting with neutrons slightly), which should happen fairly often. Much more often than the WIMPs that the dark matter experiments deep underground with billions of litres of water were designed to find.

Turns out, an experiment back in 2004 found exactly the electrical blips they were expecting to come from Axions, but the experimenters at the time dismissed the unknown blips as errors in the measurement equipment. The number of Axions, despite them being very light, could make up the mass we expect to find for 'dark matter' in the universe.


Never heard of it. I'll have to look it up!


Post new topic  Reply to topic  [ 25 posts ]  Previous  1  2



Who is online

Users browsing this forum: No registered users and 1 guest




 
     
All logos and trademarks in this site are property of their respective owner.
The comments are property of their posters, all the rest © Canadaka.net. Powered by © phpBB.